Investigate alternatives to coveralls.io and more meaningful test coverage metric

[dkaslovsky]

Builds have been failing due to outages with coveralls.io. We should investigate alternative services. Related, the test coverage metric reported for textnote is flawed as it does not discriminate between coverage of meaningful code and code that is not intended to be tested. This is always an issue with reporting test coverage but better approaches to this metric could be possible.